Integrating quantitative imaging and computational modeling to predict the spatiotemporal distribution of ~(186)Re nanoliposomes for recurrent glioblastoma treatment

摘要

Glioblastoma multiforme is the most common and deadly form of primary brain cancer. Even with aggressive treatment consisting of surgical resection, chemotherapy, and external beam radiation therapy, response rates remain poor. In an attempt to improve outcomes, investigators have developed nanoliposomes loaded with ~(186)Re, which are capable of delivering a large dose (< 1000 Gy) of highly localized β- radiation to the tumor, with minimal exposure to healthy brain tissue. Additionally, ~(186)Re also emits gamma radiation (137 keV) so that it's spatio-temporal distribution can be tracked through single photon emission computed tomography. Planning the delivery of these particles is challenging, especially in cases where the tumor borders the ventricles or previous resection cavities. To address this issue, we are developing a finite element model of convection enhanced delivery for nanoliposome carriers of radiotherapeutics. The model is patient specific, informed by each individual's own diffusion-weighted and contrast-enhanced magnetic resonance imaging data. The model is then calibrated to single photon emission computed tomography data, acquired at multiple time points mid- and post-infusion, and validation is performed by comparing model predictions to imaging measurements obtained at future time points. After initial calibration to a one SPECT image, the model is capable of recapitulating the distribution volume of RNL with a DICE coefficient of 0.88 and a PCC of 0.80. We also demonstrate evidence of restricted flow due to large nanoparticle size in comparison to interstitial pore size.